Journal of Fluids and Structures (1990) 4, 345-370

AERODYNAMI C SOURCES OF ACOUSTIC

RESONANCE IN A DUCT WITH BAFFLES
K. HOURIGAN, M. C. WELSH AND M. C. THOMPSON
Commonwealth Scientific and Industrial Research Organisation,
Division of Building, Construction and Engineering,
Highett, Victoria, 3190, Australia
AND
A. N. STOKES
Commonwealth Scientific and Industrial Research Organisation,
Division of Mathematics and Statistics,
Clayton, Victoria, 3168, Australia
(Received 21 June 1989 and in revised form 12 January 1990)
Experimental and numerical investigations of the generation of resonant sound by flow in
a duct containing two sets of baffles and the "feedback" of the sound on the vortex
shedding process are reported. The experiments are conducted in a wind tunnel and the
numerical simulations are used to predict the sources of resonant sound in the flow. The
resonant sound field, which is principally longitudinal, is calculated by the finite element
method and a discrete-vortex model is used to predict the observed separated flow.
Analysis of the passage of a single point vortex past a baffle indicates that the amount of
acoustic energy generated is a function of the phase of the acoustic cycle at which the
vortex passes the baffle. A more elaborate model simulates the growth of vortex clouds
through the clustering of elemental vortices shed from an upstream baffle, tracks the
passage of these vortex clouds past a downstream baffle, predicts the generation of
acoustic energy using Howe' s theory of aerodynamic sound, and accounts for the feedback
of sound on the vortex shedding. Comparison is made between the predicted time-
dependent structures and the observed flow structures using smoke visualization. The
vortex cloud model predicts the flow conditions under which net acoustic energy is
generated by the flow and therefore when resonance can be sustained; the results are
consistent with the occurrence of peaks in the observed resonant sound pressure levels.
1. I NTRODUCTI ON
THE EXCITATION OF ACOUSTIC MODES IN THE combust i on chamber of solid-propellant
rocket motors can lead to uncontrolled burning and vibration probl ems (Hegde &
Strahle 1985). Large flange-type structures inserted bet ween segments of grains can
result in strong shear layers which interact with similar structures downst ream,
resulting in a self-sustained oscillating system (Brown et al. 1981).
An experimental investigation of the acoustic oscillations sustained by a flow t hrough
a duct with two sets of baffles, representing an idealized solid-propellant combust i on
chamber, has been carried out by Nomot o & Culick (1982). It was found that pure
acoustic tones correspondi ng to longitudinal resonant modes of a duct were generat ed
for certain flow and geometrical conditions. When the acoustic tones were present, the
frequency of the vortex shedding from the upst ream baffles was synchronized with a
natural acoustic frequency of the duct. However, acoustic feedback was dismissed as a
means of inducing the observed vortex shedding.
0889-9746/90/040345 + 26 $03.00
C
~

V
a
r
i
a
b
l
e

s
p
e
e
d

W
i
d
e

D
.
C
.
e
l
e
c
t
r
i
c

a
n
g
l
e
d

m
o
t
o
r

d
i
f
f
u
s
e
r

C
g

0
1

.

4
-
;

;

u

J
|
l
i
h
l
l
l

~
U
L
f~
U

4

N
y
l
o
n

6

t
o

1

s
c
r
e
e
n
s

c
o
n
t
r
a
c
t
i
o
n

&

E
~
-
~
I

I

I

/

Y

~
,
1

I

I

J

I

H
o
n
e
y
c
o
m
b

~

P
e
r
f
o
r
a
t
e
d

p
l
a
t
e
s

-
-

H
o
n
e
y
c
o
m
b

4
-
2
i
7

~
-

B
.
S
.

c
o
n
i
c
a
l

i
n
i
e
t

t

A
i
r

M
i
c
r
o
p
h
o
n
e

M
i
c
P
o
p
h
o
n
e

/

~
'
~
'
+
-
t

I

-
'
'
'
.
+

1
0
9
0

I

!

I

i

J
1
3
0

=

6
1

+

2
8
6
5

(
a
)

2
4
4

A
i
r
-
-
~

v

I
:
m

o

c

>

z

N

B
a
f
f
l
e
s

-

"
/
/
/
/
/
/
/
/
/
/
/
/
/
I
/
/
/
/
/
/
/
/
/
/

F
l
o
w

~
7
/
/
/
/
/
/
/
/
/
/

V
l
l
l
l
/
/
/
/
/
/
/
/
'
f
Z
/
f
f
I
I
I
I
I
Z

-
1

I
I
I

I

L
l
/
/
l
l
/
/
/
/
/
/
/
/
/
/
/
/
/
/
~

.
.
.
.
.
.

"
~
l
I
/
/
l
l
l
l
l
l
l
l
l
l
l
~
f
/
l
I
/
l
l

(
b
)

D
u
c
t

w
a
l
l

F
i
g
u
r
e

1
.

(
a
)

S
c
h
e
m
a
t
i
c

o
f

t
h
e

t
e
s
t

r
i
g
;

(
b
)

s
c
h
e
m
a
t
i
c

o
f

t
h
e

w
o
r
k
i
n
g

s
e
c
t
i
o
n

s
h
o
w
i
n
g

r
e
g
i
o
n

o
f

f
l
o
w

v
i
s
u
a
l
i
s
a
t
i
o
n

(
b
r
o
k
e
n

l
i
n
e
s
)

c
o
r
r
e
s
p
o
n
d
i
n
g

t
o

p
h
o
t
o
g
r
a
p
h
e
d

r
e
g
i
o
n

i
n

F
i
g
u
r
e

6
.

>

0

>

348 K. HOURIGAN ET AL.
The ai m of this paper is t o descri be t he r es onant acoust i c sour ces in a duct cont ai ni ng
baffles (in t er ms of t he flow and t he acoust i c field near t he baffles) using Howe ' s (1975)
t heor y of aer odynami c sound. Fi gur e 1 shows schemat i cs of t he t est rig and wor ki ng
sect i on. Fur t he r mor e , t he paper aims t o est abl i sh whet her acoust i c f eedback is t he
mechani sm whi ch synchr oni zes t he vor t ex sheddi ng. The backgr ound t o t he pr es ent
st udy is pr es ent ed in Sect i on 2. Sect i on 3 descr i bes t he exper i ment al appar at us and t he
pr ocedur es used f or measur i ng t he r es onant acoust i c sound and phot ogr aphi ng t he
flow. Descr i pt i ons of t he finite el ement me t hod f or pr edi ct i ng t he acoust i c field, t wo
flow model s and t he aer oacoust i c t heor y of Ho we (1975) r el evant t o t he pr obl em ar e
pr es ent ed in Sect i on 4. Sect i ons 5 and 6 cover t he resul t s and discussion per t ai ni ng t o
bot h t he exper i ment al and t he comput at i onal i nvest i gat i ons. The pr edi ct ed r es onant
acoust i c field and t he acoust i c St r ouhal number s at whi ch acoust i c r es onance is
obser ved ar e pr esent ed. Onl y t he case f or t he si mpl est acoust i c mode whi ch can be
exci t ed, cor r espondi ng t o t hat wi t h a wavel engt h appr oxi mat el y equal t o t he l engt h of
t he wor ki ng sect i on, is descr i bed. To i l l ust rat e qual i t at i vel y t he sound gener at i on
mechani sm, a simplified flow model wi t h a single vor t ex passing a baffle is pr es ent ed.
Thi s model i ndi cat es how t he flow near t he downs t r eam baffle can act as an acoust i c
sour ce r egi on, pr ovi ded t he vor t i ces pass t he baffle at t he appr opr i at e phase in t he
sound cycle. In t he second model , t he flow is r epr es ent ed as t he sum of a st eady
pot ent i al flow, a flow wi t h vor t ex cl ouds in f r ee mot i on, and an oscillating flow
r epr esent i ng t he acoust i c field. The vor t ex cl oud model addi t i onal l y account s f or t he
f eedback of t he r esonant sound on t he vor t ex sheddi ng f r om t he ups t r eam baffle and
predi ct s t he acoust i c St r ouhal number s at whi ch peaks occur in t he acoust i c ener gy per
cycle gener at ed by t he flow. Indi vi dual vor t ex st r uct ur es are also t r acked t o show t hei r
rol e in sustaining t he acoust i c r esonance at f avour abl e acoustic St r ouhal number s. The
pr edi ct i ons are compar ed wi t h exper i ment al results. Sect i on 7 sets out t he concl usi ons
of t he st udy.
The maj or resul t of t he st udy is t hat a f eedback l oop can be est abl i shed bet ween a
r esonant sound field and t he vor t ex sheddi ng f r om a set of baffles in a duct . Al t hough
this l oop can be est abl i shed over a numbe r of flow vel oci t y ranges, t he c ommon f act or
l eadi ng t o t he mai nt ai ni ng of t he l oop is t he phase of t he acoust i c cycl e when each
vor t ex arri ves at t he downs t r eam baffles. It is onl y f or par t i cul ar r anges of flow
velocities t hat t he phasi ng bet ween t he vor t ex passi ng t he downs t r eam baffle and t he
r esonant acoust i c field is such t hat sufficient posi t i ve net acoustic ener gy is gener at ed t o
sustain t he acoust i c r es onance and t he f eedback l oop.
2. BACKGROUND
Fl ow i nduced acoust i c r esonances in duct s can mani f est in di fferent modes dependi ng
on vari ous fact ors, i ncl udi ng duct modi f i cat i on and flow vel oci t y. For si mpl e r ect angu-
lar duct s, t he modes can be classified as ei t her t r ansver se or l ongi t udi nal , dependi ng on
t he gener al di r ect i on of t he wave mot i ons. Single or mul t i pl e sour ces of sound ar e
f ound f or t he t r ansver se modes; t he maj or sour ce of sound may or may not be r e mot e
f r om t he poi nt of vor t ex sheddi ng t o whi ch t he sound f eeds back and cont r ol s.
Gener al l y, l ongi t udi nal modes are exci t ed by acoust i c sources r e mot e f r om t he poi nt of
vor t ex sheddi ng.
A br i ef revi ew of some pr evi ous i nvest i gat i ons i nvol vi ng di fferent modes and sour ce
regi ons is pr es ent ed in this sect i on.
ACOUSTIC SOURCES IN A DUCT WITH BAFFLES 349
2.1. EXCITATION OF TRANSVERSE DUCT Ac ous Ti c MODES
Parker (1966) and Parker & Grifiiths (1968) made detailed studies of acoustic
resonances induced by vortex shedding from vertical cascades of plates located in
horizontal flow ducts; the resonances were excited purely by the flow and did not
require structural vibration for their excitation (Parker 1969). Parker (1967) also
predicted the resonant frequencies by solving the wave equation for the space
surrounding the plates. He showed that the frequencies depend on the dimensions of
the plate as well as the dimensions of the duct. The simplest acoustic mode was defined
to be the fl-mode which describes the acoustic field both near the plate and in the far
field, where it becomes an evanescent cross-mode of the duct. Later studies of acoustic
resonance by Cumpsty & Whitehead (1971), Archibald (1975), Welsh & Gibson (1979)
and others observed resonances when a single plate was installed in a duct.
2.1.1 Acoustic source coincident with vortex shedding region
Welsh et al. (1984) reported a more detailed study of the excitation of acoustic
resonances in flow around plates with semicircular leading edges. They examined the
fluid mechanics of the resonant process in a duct containing a plate with a semicircular
leading edge. The process is described in terms of an interchange of energy between
the flow and acoustic fields and has three basic components: (a) an acoustic source (the
vortex street); (b) a feedback effect of the sound on the vortex shedding; and (c) a
damping process whereby acoustic energy is transferred out of the duct system. This
study showed that there was an acoustic source downstream of the trailing edge where
vigorous vortex shedding occurred.
2.1.2. Acoustic source remote f r om vortex shedding region
Single square leading edge plates
Welsh & Gibson (1979) examined the interaction between resonant sound and
vortex shedding from a flat plate with a square leading edge. They found that only
those acoustic modes with acoustic velocity antinodes near the plate and the associated
vortex street were excited by the vortex shedding. They also found that plates with
square leading edges could shed large-scale vortices from either the leading edge or the
trailing edge during resonance, and showed that the same acoustic mode could be
excited at the same frequency over several ranges of flow velocity.
Stokes & Welsh (1986) modelled the acoustic resonance process for the case of
plates with square leading edges. The mathematical model showed that in resonance,
the vortices observed shedding from the leading edge during each acoustic cycle
generated a net supply of acoustic energy as they passed the trailing edge, provided the
passage was at a particular phase of the sound cycle. A variety of solutions was found
to exist for the number of sound cycles that may pass while a vortex travels from the
leading edge to the trailing edge so that it arrives there at the appropriate phase in the
sound cycle. Correspondingly, there are several ranges of flow velocity at which this
phase can be achieved to generate a net supply of acoustic energy. These predicted
flow velocity ranges closely coincided with the velocity ranges for which acoustic
resonance is observed in practice (Stokes & Welsh 1986). Since a vortex is shed each
acoustic cycle during resonance, each velocity range corresponds to a different number
of vortices observed at a particular instant between the leading and the trailing edges.
350 K. HOUR I GAN ET AL.
More recently, Welsh et al (1990) have observed that, similar to the case of square
leading edge plates, vortices are shed from semicircular leading edges of plates, and
that this vortex shedding can excite, and in turn be synchronised by, a transverse duct
mode.
In all of these cases, a strong acoustic source was present downstream of the region
where vortex shedding was synchronized by the resonant sound.
Mul t i pl e pl at es
If two plates with semicircular leading edges are placed in tandem in a duct, a
resonant transverse acoustic mode of the duct can be excited, leading to feedback of
the sound onto the separating shear layers at the trailing edge of the upstream plate
(Stoneman et al. 1988). The amplitude of the sound can be substantial, resulting in the
locking of the vortex shedding to the resonant sound frequency. The periodic
reappearance of loud acoustic resonances as the plate spacing was increased is similar
to that observed for single plates with blunt leading edges as the flow velocity was
decreased. In both cases, resonance relied on the vortices shed upstream passing either
a trailing edge or a second plate at an appropriate phase of the acoustic cycle.
2.2. EXCITATION OF LONGITUDINAL DUCT ACOUSTIC MODES
Nomoto & Culick (1982) published data showing vortex shedding exciting acoustic
resonances in rectangular ducts containing two pairs of baffles; as distinct from
transverse resonance modes investigated by Stokes & Welsh (1986) and Stoneman et
al. (1988), the resonances were essentially the longitudinal organ pipe modes of the
duct. As for the case described above of the plate with a square leading edge, Nomoto
& Culick (1982) observed the same resonant acoustic mode being excited over several
different ranges of flow velocity. Again, each range was uniquely characterized by the
number of vortices observed between the baffles at a given instant. Harris et al. (1988)
investigated the air flow through a double orifice in a pipeline; acoustic resonance in
the pipe mode could be excited at particular flow velocities, and shear layer oscillations
leading to the formation of large-scale vortex structures at the resonant frequency were
observed.
Some flow geometries, such as jet-edge interactions, lead to self-sustaining oscilla-
tions by purely hydrodynamic perturbations (Rockwell 1983); the accompanying sound
produced by the flow oscillations is not amplified by duct reflections. However, the
frequencies of the oscillations and consequent sound are functions of the flow velocity,
whereas in the present case, the vortex shedding and sound frequencies are
predominantly a function of the duct geometry and involve loud resonant sound levels.
This suggests that for the current investigation, the influence of sound on the
development of flow structures is important and that the mechanism leading to the
oscillation is fundamentally different to that proposed by Rockwell (1983) for a
different flow. Nomoto & Culick (1982) cast doubt on the influence of resonant sound
on the vortex shedding process for flow in a duct geometry similar to the present
investigation, in which the flow is reexamined experimentally and theoretically. Chung
& Sohn (1986) and Flandro (1986) investigated the coupling mechanism of acoustic and
vortical wave oscillations in connection with rocket combustion chambers using linear
stability theory. Their results, which are relevant only to the initial excitation of an
acoustic resonance, clearly indicate that a feedback loop can be established between
the sound field and the growth of instabilities in the shear layer. The present work,
ACOUS T I C S OURCE S IN A DUCT WI TH BAF F L E S 351
following Stokes & Welsh (1986) and Stoneman et al. (1988), is concerned with the
non-linear coupling of finite resonant acoustic waves and vortex shedding which is
observed when a loud acoustic resonance has been excited.
2.3. DETERMINATION OF THE ACOUSTIC SOURCES
In the studies of Welsh et al . (1984), Stokes & Welsh (1986) and Stoneman et al .
(1988), and in the present study, the source of acoustic energy due to the vortex street
is modelled using Howe' s (1975, 1980) theory of aerodynamic sound. According to this
theory, the acoustic power generated by a vortex as it passes through a sound field is
proportional to the scalar triple product of the vorticity, the velocity of the vortex and
the acoustic particle velocity. For low Mach number flows the acoustic and flow fields
can be decoupled and solved for separately. Then, assuming a resonant sound field is
present, the acoustic power generated by a vortex passing through it can be computed
using Howe' s theory. If, on integrating the power over time, there is a net source of
acoustic energy then resonance is possible, provided damping is overcome. However, a
net negative source of energy implies that a resonance is not sustainable.
It is the purpose of this paper to show that the acoustic sources in a flow around
baffles in a duct can be examined in a manner analogous to the flows around plates,
even though the excited resonant acoustic mode is longitudinal rather than transverse.
3. EXPERIMENTAL APPARATUS AND PROCEDURES
The experimental apparatus was a small blow down wind tunnel, shown schematically
in Figure l(a). A British Standard conical inlet 217 mm in diameter was connected to
the fan inlet for monitoring the volume flow through the tunnel. A wide angle diffuser
containing four perforated plates connected the fan to the settling chamber which
consisted of a honeycomb and four fine nylon screens. From the settling chamber, air
passed through an 8 to 1 contraction before entering the working section, which
consisted of rigid ducting with internal dimensions of 244 mm square. The total length
of the 244 mm square ducting was 2865 mm. Some portions of the modules were
constructed from 12mm thick aluminium and contained facilities for mounting
microphones flush with the inside surface of the duct, while other parts of the ducting
were constructed from 25 mm and 50 mm thick acrylic (methyl-methylacrylate) sheets
with optical glass inserts for flow visualization (Figure lb). A chamber 130 mm long in
the flow direction was formed midway along the ducting by installing two pairs of
baffles across the flow inside the ducting. The opening between the baffles was 61 mm,
which was 25% of the full duct width.
The velocity profile approaching the baffles was uniform (within 1%) between the
wall boundary layers, which were approximately 15 mm thick. The axial component of
the turbulence intensity in the core of the flow was 0.3% with significant spectral
content less than 100 Hz.
Acoustic resonances were identified by recording the signal from a fixed 12.7 mm
diameter Briiel and Kjaer microphone mounted flush with the surface inside the
chamber (Figure 1). This microphone was located near a duct corner in a plane
116.5 mm downstream of the upstream baffle. A second Briiel and Kjaer microphone
was located near a duct corner and 285 mm upstream of the upstream baffle. The flow
and acoustic variables were recorded for the maximum sound pressure level for each
resonance as the mean flow velocity upstream of the baffles was increased from zero to
352 ,K. HOURI GAN ET AL.
5 m/s. The resonance frequencies were obtained by performing a ninth order fast
Fourier transform with ten averages on the microphone data.
The sound pressure levels at the acoustic resonance frequency were determined from
narrow band analysis (bin width of 1-95 Hz) of the microphone signals. Since these
signals were tonal and up to 40 dB above the background, the amplitude of the spectral
peak approximated the sound pressure level at the required frequency. The amplitude
of the spectral peak was calibrated using an input signal of known sound pressure level
from a Briiel and Kjaer piston phone.
The phase relationship between the vortices and the acoustic field was determined by
introducing smoke into the working section upstream of the baffles and photographing
the flow at selected points in the acoustic cycle using a time delay unit connected to the
microphone signal. The flash photographs shown in this paper were obtained by
superimposing ten exposures taken for the same flow conditions, 0.16 +0.03 of an
acoustic cycle before the minimum acoustic pressure was reached at the microphone
upstream of the baffles [Figure l(a)].
4. THEORY AND MATHEMATI CAL MODELLING
In the following mathematical models, the ducts and baffle geometries are based on the
test rig shown in Figure l(a). As a result of the symmetry of the flow and the acoustic
field about the mid-line of the duct, only one half of the duct flow is explicitly
modelled.
4.1. ACOUSTIC FIELD
The acoustic mode to be modelled is a standing wave corresponding to an organ pipe
mode. In the flows of interest here, the Mach number is small and the acoustic pressure
p satisfies the wave equation:
32P 2 2
3. g2 m. C V p (1)
where c denotes the velocity of sound and z is time.
The time-dependent amplitude function q~ can be extracted from a standing wave
solution p -- q~e i2~#, where f is the frequency. Then tp satisfies the Helmholtz equation:
V2tp + (2a~f/c)2q~ = O. (2)
For the low frequency modes, which are symmetrical about the horizontal midplane,
the following boundary condition applies on the rigid surfaces and also on the
midplane:
k. V = 0, (3)
where k is the unit vector normal to the surface. The boundary conditions at the ends
of the working section of the duct are, in reality, more complex due to the presence of
other surfaces in the wind tunnel. However, as Blevins (1985) has also found, the
solutions in the inner region of the duct where the baffles are located are insensitive to
whether Dirichlet or Neumann conditions are used for the end boundary pressures. In
the present study, the Dirichlet condition at the ends, i.e. p = 0, is employed.
For the simplified single vortex model of the flow, the sound field, which has a long
wavelength compared with the dimensions of the baffles, is approximated locally by the
ACOUSTI C SOURCES IN A DUCT WITH BAFFLES 353
Fi gur e 2. Pa r t o f t he me s h u s e d f or fi ni t e e l e me n t pr e di c t i on o f t he a c ous t i c field.
potential flow solution around a baffle det ermi ned using the conformal transformations
described below in Section 4.2.1.
For use in the vortex cloud model , the boundary value probl em is solved by t he finite
element met hod using rectangular elements and piecewise linear shape functions. The
orthogonal mesh used, shown in Figure 2, is compressed near t he tips of the baffles and
is generated using a finite difference met hod solved by line successive over-relaxation.
The met hod, due to Warsi (1982), is based on t he idea that different coordinate
systems in a given region should be related by the transformation laws of tensors. The
lowest frequency acoustic mode is found by an iteration procedure in which the
eigenvalue is estimated at each step by t he Rayleigh quot i ent (e.g. Zienkiewicz 1978).
Successive higher modes can then be found in a similar manner.
4.2. FLOW MODELLING
4.2.1. Single vortex model
This model is used to investigate the resonant acoustic power generat ed by a single
point vortex passing an isolated baffle in duct. The model assumes a two-dimensional
inviscid incompressible flow, irrotational everywhere except at the centre of t he point
vortex. For simplicity, the vortex of infinitesimal strength is assumed to follow a
horizontal path with velocity det ermi ned by the horizontal component s of the
irrotational flow velocity and the induced velocity due to the image vortex. Since the
working section is long compared with the flow features, its length is assumed infinite.
The transformation which takes the complex potential w for a flow past a set of thin
baffles into the variable z, which represents the points in the physical plane, is given by
( c s a t
z = - log \ c o s a + v ' (4)
where v = (tanh 2 w- sin 2 a) , a is the baffle height and d is the spacing bet ween the
duct walls.
4.2.2. Vortex cloud model
In this model, the flow is modelled by a two-dimensional inviscid incompressible flow,
irrotational everywhere except at the centres of el ement al vortices. Due to the
symmetry of the flow and the resonant acoustic field about t he longitudinal centreline
of the duct, only the flow in the lower half duct is explicitly modelled. The shedding of
vorticity is modelled by the creation of the elemental vortices, which are convect ed
under the influence of ot her elemental vortices and the irrotational flow.
In the single vortex model , no attempt was made to simulate the process of vortex
formation through instability in separating shear layers. However , in this section, the
vortical flow around t he baffles in the duct, which includes the instability of the
separating shearlayer, is model l ed by releasing elemental vortices from the surfaces of
354 K. HOURIGAN ET AL.
the baffles and tracking their motions as t hey evolve into clouds. To satisfy the rigid
duct wall condition, a similar mapping t o that empl oyed for the single vort ex
calculation described in Section 4.2.1 was used. This mapping w = e z t ransformed the
duct into a half-plane ().-plane) and t he vortex-induced pert urbat i ons were forced to
meet the duct-wall conditions by introducing mirror images. The no slip boundar y
condition was satisfied at the baffle surfaces through the surface vorticity met hod
(Lewis 1981). Figure 3 shows the baffle geomet ry in the original z-plane and
transformed ).-plane.
The surface vorticity met hod has been used previously by Lewis (1981) and
Stoneman et al. (1988), amongst others, t o represent the surface of a bluff body by a
vortex sheet. In brief, it is required that t he cont our along the body surface is a
streamline and that the tangential velocity on the inside of t he vort ex sheet is zero.
Denoting the distance along the body surface by s, discretization of a vort ex sheet into
M segments, with the nt h segment having length As, and linear vorticity density y( s , ) ,
produces a set of linear equations,
M U I d X m \ I d y m \
Z ~ ( s , ) g ( s n , S m) mS , - 1 7(Sm) = --(Ux Of. x , a ) ~ . . ~ ) _ _ ( ~ j y - 1 - Uy , a ) ~. - ~)
n=l
N,
- ~ F . T(n, sin), (m = 1, M), ( 5 )
n=l
where t he last t erm in equation (5) gives the contribution to the velocity field at the
surface due to No free vortices of circulation F in the flow. The coupling coefficient
K( s, , sin) has the value of the surface tangential velocity at Sm induced by a vort ex of
unit circulation at s, . The coupling coefficient T(n, s , ) has the value of the surface
z - pl ane
B A
R R
C D
X - pl ane
A' B' C' D'
Figure 3. Transformation used in the vortex cloud model mapping the half duct to a half plane together
with the baffle geometries in each plane.
ACOUSTIC SOURCES IN A DUCT WITH BAFFLES 355
tangential velocity at Sm due to a vortex of unit circulation at the position of the nth
free vortex. The velocity of the irrotational flow has been separated into a steady
component ( V x , V y ) and an unsteady component ( v ~ . a , V y . a ) contributed by the acoustic
field; owing to the long wavelength of the acoustic field relative to the dimensions of
the vortex structures, the flow due to the acoustic field is almost potential locally in the
region of the baffles and is incorporated into the potential flow. The solution of (5)
gives the surface vorticity density at the pivotal points on the surface of the body,
which are taken to be the centre of each discrete element.
In the scheme used by Stoneman e t a l . (1988), elemental vortices were released into
the flow from a very limited number of designated flow separation points. In the
present work, this condition has been relaxed, in a manner similar to that of Smith &
Stansby (1988). At each time step, the elemental vortices are created having the
circulation of each vortex segment on the surface of a baffle; vortices on the upstream
faces of the baffles are released into the flow to simulate the formation of the boundary
layers. Interactions between the elemental vortices in the recirculating flow in the
cavity region between the baffles can have an unphysically disruptive influence on the
sensitive separating shear layer unless very large numbers of vortices are employed. As
this was not computationally feasible in the present study, the time-mean vorticity in
the cavity region predicted from an initial simulation was fixed on a 21 x 21 grid. This
representation provided the approximately correct velocity contribution to the shear
layer under consideration from the cavity region without introducing instabilities due to
numerical noise; satisfactory vortex development was then predicted in the shear layer
separating from the upstream baffle.
The elemental vortices are potential vortices with smoothed cores (Rankine profile)
of radius 0-03b, where b is the spacing between the upstream and downstream baffles.
Test cases with larger and smaller smoothing cores showed that the formation of the
large-scale vortex structures was insensitive to the exact smoothing value used. The
vortices are advected using a second-order Adams-Bashford scheme; a time-step of
O . O l ( b / o ~ o ) is used, where v~ is the upstream flow velocity. As mentioned above,
prediction of the level of the resonant acoustic particle velocity amplitudes was not
attempted; these amplitudes were determined from the empirical data. For input to the
numerical model, the value for the amplitude of the acoustic particle velocity at the
centre of the duct midway between the baffles is set to 0-1v, which is in the middle of
the empirical range. Amplitudes fifty per cent higher and lower than this value were
also used in the model to test the sensitivity of the results; near identical results were
obtained. To economize on the number of computations required, the vortices are
merged according to an exponential weighting function of the distance downstream of
the upstream baffle.
4. 3. INTERACTION OF FLOW AND SOUND
Howe (1975, 1980) showed that when an acoustic oscillation occurs in an inviscid,
isentropic but rotational flow, then an instantaneous acoustic power P is generated in a
volume V, which is given by
P = - p o l O S " (v X u) dV (6)
where v is the fluid velocity, ta = V v is the vorticity, u is the acoustic particle velocity
and/9o is the mean density of the fluid.
356 K. HOURIGAN ET AL.
Whe n t he vort i ci t y is " c ompa c t " , t hat is when t he vort i ci t y ext ends over a r egi on
whi ch is small rel at i ve t o t he acoust i c wavel engt h, t hen t he acoust i c power per uni t
l engt h of vor t ex t ube gener at ed by a vor t ex r educes t o
P = p o r k . (v x u) , (7)
wher e r is t he ci rcul at i on of t he vor t ex and k is t he uni t vect or nor mal t o t he pl ane of
t he flow.
For t he pr esent anal ysi s, wher e t he fl ow near t he downs t r eam baffle will be i dent i fi ed
as an acoust i c sour ce, it is useful t o r ewr i t e t he above expr essi on f or acoust i c power P
in t er ms of t he phase of arri val of t he vor t ex at t hat baffle r el at i ve t o t he sound cycl e. P
is osci l l at ory over t i me because of t he osci l l at i on of t he acoust i c par t i cl e vel oci t y; t o
separ at e this osci l l at ory t er m, first u is f act or i zed:
n = Uo sin[2~rf(T - ~:0) + h] , (8)
wher e Uo is t he st eady ampl i t ude of t he s ound field, f is its f r equency and To is t he t i me
at whi ch t he vor t ex is di rect l y above t he downs t r eam baffle. He r e , h is t he phase of t he
acoust i c cycle at whi ch t he vor t ex passes over t he downs t r eam baffle. The n
P = Q sin[2~rf(~" - T0) + h], (9)
wher e
a = - p o r Ivl In01 sin e (10)
and e is t he angl e bet ween v and u0.
Two si mpl e exampl es will ser ve t o expl ai n t he physics of equat i ons (9) and (10).
First, if a vor t ex t ravel s parallel t o t he local acoust i c part i cl e vel oci t y ampl i t ude a0,
t hen t he angl e e bet ween t he vor t ex vel oci t y v and Uo is zer o. The r e f or e , Q in (10) is
zer o and accor di ng t o (9), no i nst ant aneous acoust i c power P is gener at ed. Conse-
quent l y, t he net acoust i c ener gy gener at ed over an acoust i c cycl e is zer o. Thi s si t uat i on
arises, f or exampl e, i f t he vor t ex is t ravel l i ng al ong a duct in a l ongi t udi nal st andi ng
wave f ar away f r om duct modi fi cat i ons. Second, if a vor t ex t ravel s orthogonal t o t he
local acoust i c part i cl e vel oci t y (i . e. = + ~/ 2) , t he ampl i t ude of Q is maxi mum and
t he i nst ant aneous acoust i c ener gy P oscillates wi t h t i me: I f t he vor t ex vel oci t y and t he
local acoust i c par t i cl e vel oci t y ampl i t ude r emai n const ant , t hen t he i nt egr at ed ener gy
over an acoust i c cycle is zer o. The r e f or e , f or net acoust i c ener gy t o be gener at ed over a
cycle, Q must vary, whi ch occurs most r eadi l y near a duct modi fi cat i on (e. g. a baffle)
wher e t he angl e bet ween t he local acoust i c par t i cl e vel oci t y and t he vor t ex vel oci t y can
var y r api dl y in space. In t he pr esent i nvest i gat i on, a net acoust i c ener gy gener at ed over
a cycle will be shown t o be possi bl e when a vor t ex passes t he downs t r eam baffle.
The sign and amount of acoust i c ener gy gener at ed by vor t ex st r uct ur es will be shown
t o be critically dependent on t he phase h of t he acoust i c cycle at t he i nst ant of arri val of
a vor t ex st r uct ur e at t he downs t r eam baffle. Tha t is, t he phase offset h plays a cruci al
r ol e in t he aer oacoust i c process. Some key val ues of h are: h = 0, t he r es onant acoust i c
part i cl e vel oci t i es are zer o ever ywher e and changi ng t o a di r ect i on whi ch augment t he
mean flow; h = ~r/2, t he acoust i c par t i cl e vel oci t i es ar e maxi mum and in t he flow
di rect i on; h = ~r, t he acoust i c part i cl e vel oci t i es ar e zer o ever ywher e and begi nni ng t o
oppose t he mean flow; h = 3/2~r, t he acoust i c par t i cl e vel oci t i es are maxi mum in t he
upst r eam di rect i on.
The exci t at i on of an acoust i c r es onance f r om an initially l ow l evel is a compl ex
t r ansi ent process. Li near analysis of an initial exci t at i on of a r es onant acoust i c mode
has been pe r f or me d by Chung & Sohn (1986) and Fl andr o (1986). In t he pr es ent wor k,
at t ent i on is focussed on whet her a r es onance can be sust ai ned at t he est abl i shed
ACOUSTIC SOURCES IN A DUCT WITH BAFFLES 357
observed levels and the nature of the dominant sources of sound in the flow that can
sustain the resonance. A resonant sound field will be assumed to be present, and in
some cases the predicted mean acoustic energy per cycle generated by the flow will be
found to be relatively low or negative. Although the vortiCal structures that evolve as a
result of the assumed resonance in these cases are fictitious, the result shows that such
a resonance cannot exist and describes potential acoustic sinks. To determine whether
the predicted acoustic energy per cycle generated by the flow is consistent with
observed resonances at various flow velocities, the following strategy is adopted.
(i) The principally longitudinal acoustic resonant mode is predicted using the finite
element technique.
(ii) The predicted resonant acoustic field at the observed sound pressure level is
incorporated in the flow model, resulting in the synchronized shedding of large-scale
vortices from the leading baffle.
(iii) The acoustic energy per cycle generated by the entire flow is then calculated.
(iv) If the acoustic energy per cycle generated by the flow is relatively low, then
resonance is deemed not to be sustainable.
(v) This procedure is repeated for a number of different flow velocities. The flow
velocities at which peaks in the generated acoustic energy per cycle are maxima are
compared with those velocities at which resonances are observed.
(vi) The instantaneous acoustic power generated by a single large-scale vortex
during its passage past the downstream baffle is predicted; this provides insight into
the source of the acoustic resonance and why the resonance is observed for only certain
discrete velocity ranges.
5. RESULTS
5.1. PREDICTED RESONANT ACOUSTIC FIELD
The local acoustic particle velocity magnitudes, predicted by the finite element method
(Section 4.1) for the resonant longitudinal field (having a wavelength approximately
equal to the length of the duct) with baffles installed (Section 4.1), are shown in Figure
4(a). The velocity directions are generally horizontal except near the baffles where
significant vertical components are found. The acoustic particle "streamlines" calcu-
lated using the potential flow solution (Section 4.2.1) approximation are shown in
Figure 4(b).
5. 2. EXPERIMENTALLY OBSERVED RESONANCES
The relationship between resonance and flow velocity is most clearly shown in terms of
an acoustic Strouhal number St,. This is formed by multiplying the sound frequency f
by the distance between the baffles b (130mm) and dividing this product by the
average velocity upstream of the baffles. The reference length is chosen to be the
distance between the baffles rather than the acoustic wavelength because it will be
shown that the time taken by a vortex to traverse this spacing determines whether the
acoustic resonance can be sustained.
The power spectra of the signal detected by the microphone located between the
baffles when resonance occurs are shown in Figure 5. Frequency, which is normally
plotted on the horizontal axis, is replaced by the acoustic Strouhal number Sta. Peaks
in the spectra at acoustic Strouhal numbers of 11.0, 7.7 and 4.7 indicate the amplitude
of the resonances which are at the same frequency ( = 100 Hz) and are the resonant
358 K. HOURI OAN ET AL.
I T I I I I
I ,lla
\ / - -
/ _
/ ~ ~.L I
~ ---"-I P. "~ _
...,.,,.
"" "7, r'l ",- - / _
' ' H ," -
' ; H
I 1 : - H I
(a)
D u c t c e n t r e l i n e
. . . . . . . . . . . .L
(b} B a f f l e D u c ! wall
Fi gure 4. ( a ) Ac o u s t i c parti cl e ve l oc i t y vect ors pr e di c t e d by t he fi ni te e l e me nt me t ho d at a phas e o f t he
s ound c yc l e whe r e t he acous t i c part i cl e ve l oc i t y i s gr e at e s t i n t he downs t r e am di rect i on; ( b) predi ct ed
acous t i c part i cl e "st reaml i nes" near an i s ol at e d baffl e us e d f or t he s i ngl e vor t e x mode l .
mode descri bed in Sect i on 5. 1. They occur at different mean upst ream flow vel oci t i es
o f 1-2, 1.7 and 2-7 m/ s , respecti vel y. At the hi ghest flow vel oci t y, harmoni cs o f t he
primary peak al so appear but at consi derabl y l ower ampl i tude than the primary peak.
Three different mul ti pl e flash phot ographs of t he fl ows bet ween the baffles are s hown
in Fi gure 6 for the condi t i ons correspondi ng t o t he s ound pressure l evel peaks in the
spectra s hown in Figure 5. In each phot ograph t he resonant acousti c mo de and the
f requency is the same ( - 100 Hz) . It is clear that at acousti c Strouhal numbers o f 11. 0,
7. 7 and 4. 7, there are, respecti vel y three, t wo or one pair(s) o f vorti ces bet ween the
baffles, as f ound al so by No mo t o & Cul i ck (1982).
ACOUSTIC SOURCES IN A DUCT WITH BAFFLES 359
130 = n I = ~ t R
120
(a)
110 ( b)
100
( c)
"O
' - 8()
- i
O
~ J
70
1
6vO 2 4 6 8 10 12 14 16
Acoust i c St r ouhal number , St a
Figure 5. Plot of t he power spect rum of t he signal from t he mi crophone l ocat ed bet ween t he baffles at t he
acoustic Strouhal numbers (a) St= = 4-7; (b) Sta = 7"7; (c) St . = 11"0. Each plot has a marked peak (shaded)
at the resonant frequency; t he ot her maj or peaks shown in t he spect rum are harmonics.
(a)
(b)
(c)
Figure 6. Phot ographs using smoke visualisation of the flow for t he first t hree peaks in sound pressure
level found at t he acoustic Strouhal numbers (a) St,, = 4.7; (b) St , = 7-7; (c) St, = 11.0.
360 K. HOUR1GAN ET AL.
5.3. SINGLE POINT VORTEX MODEL
F i g u r e 7 s h o ws t h e i n s t a n t a n e o u s a c o u s t i c p o w e r g e n e r a t e d b y a v o r t e x , h a v i n g
i n f i n i t e s i ma l s t r e n g t h a n d f o l l o wi n g a h o r i z o n t a l p a t h , a s i t p a s s e s o v e r t h e d o w n s t r e a m
b a f f l e ; t h e b a f f l e a n d v o r t e x a r e b o t h s u r r o u n d e d b y a l o n g i t u d i n a l a c o u s t i c f i e l d. T h e
i n s t a n t a n e o u s p o w e r i s s h o w n f o r f o u r d i f f e r e n t v a l u e s o f t h e p h a s e o f f s e t h , wh i c h
s p e c i f i e s t h e p h a s e o f t h e a c o u s t i c f i e l d a t t h e i n s t a n t t h e v o r t e x p a s s e s o v e r t h e
d o w n s t r e a m b a f f l e [ s e e e q u a t i o n ( 8) ] . I t i s o n l y w h e n t h e v o r t e x i s n e a r , b u t n o t d i r e c t l y
a b o v e , t h e b a f f l e t h a t t h e a n g l e [ s e e e q u a t i o n ( 10) ] i s s i g n i f i c a n t a n d t h e a b s o l u t e
v a l u e o f P i s r e l a t i v e l y l a r g e . T h a t i s , t h e d i r e c t i o n o f m o t i o n o f t h e v o r t e x a n d t h e l o c a l
m m m ( a ) " - - - -
( b )
. . . . . . . . ' " l l l l l l l l ' I ' N l l l l l l l l " . . . . . . . . .
B a f f l e I
D u e t w a l l
+p
............ p
( c )
. . . . . . . . J ~ J l l l l l l l l , I I I I I l l l t l , , . . . . . . . . . . . . +
P
q
( d )
. . . . . . . . . . . . . . . . . . . . . , , 1 1
-I-
_ P
Figure 7. Plots of predicted instantaneous acoustic power P (arbiatrary units) for a single vortex passing a
baffle for four different values of the phase h: (a) h = 0; (b) h = at/2; (c) h = ~r; (d) h = 3/ 2n; I line segments
plotted orthogonal to the horizontal path of the vortex have lengths proportional to the value of P at that
point of the trajectory.
ACOUSTI C SOURCES IN A DUCT WITH BAFFLES
361
acoustic particle velocity direction are non-parallel on either side of the baffle and
acoustic power generation is possible. Far upstream and downstream, and directly
above the baffle, the two directions are parallel and no acoustic power generation
possible. The sign of the instantaneous acoustic power generated depends on the sense
of the local acoustic particle velocity, which in turn depends on the phase offset h. As
the phase offset h is varied, different combinations of positive and negative peaks in the
time history of the acoustic power generation appear. Two equal positive peaks are
found for h = ~r and two equal negative peaks are found for h = 0.
The net acoustic energy generated by a point vortex during its passage past the baffle
is a sinusoidal function of the phase offset h, under the present assumptions. This can
be easily shown by integrating equation (9) over time. The net acoustic energy reaches
its maximum value (positive) when h = ar and its minimum value (negative) when
h = 0. For h = at/2 and h = 3/2~r, the net generated acoustic energy over time is zero.
5.4. VORTEX CLOUD MODEL
From the vortex cloud model, the predicted acoustic energy output per acoustic cycle
due to the entire flow is shown in Figure 8 over a range of acoustic Strouhal numbers.
The total power was obtained by summation of the individual powers (7) over all the
vortices in the flow over ten consecutive acoustic cycles. The acoustic Strouhal numbers
at which peaks in the acoustic energy per cycle generated by the vortex clouds occur
are found to coincide well with the observed values of 7-7 and 4.7; a predicted third
peak near the observed value of 11.0 is only slightly resolved.
"Snapshots" showing the predicted elemental vortex positions are shown in Figure 9
for the acoustic Strouhal numbers corresponding to peaks in the time-mean sound
pressure level, and the same phase of the acoustic cycle, as for the photographs shown
in Figure 6. At these acoustic Strouhal numbers of 4-7, 7-7 and 11-0, there are one, two
and three pair(s) of vortices, respectively, between the baffles. There is clearly a
feedback of the resonant sound on the flow separating at the upstream set of baffles,
leading to the vortex shedding being locked in frequency to the excited resonant
acoustic mode.
Figure 10 shows the predicted instantaneous acoustic power output P generated by
the flow over a number of cycles for two acoustic Strouhal numbers, Sta--4.7 and
Sta--6-0, corresponding to a maximum and minimum, respectively, of the predicted
I I I I I
~e

E~
I I I I I
0 2 4 6 B 10 12
Acoust i c Strouhal number, Stn
Fi gur e 8. Pr edi ct ed g e n e r a t e d acous t i c e ne r gy p e r cycl e ( ar bi t r ar y uni t s ) d u e t o t he ent i r e fl ow de r i ve d
f r om t he vor t e x cl oud mo d e l ove r a r a nge of acous t i c St r ouha l n u mb e r s .
362 K. HOURI GAN ET AL.
(a) .
~b)
Figure 9. Predicted vortex positions (superimposed from the same phase of the acoustic cycle over five
consecutive acoustic cycles) corresponding to Figure 6 for the three acoustic Strouhal numbers associated
with peak acoustic resonances: (a) St~ = 4.0; (b) St. = 7-7; (c) Sta = 11.0. Elemental vortex positions are at
centres of circles.
acoustic energy per cycle generat ed. The average generat ed acoustic energy per cycle is
found t o be negat i ve when St, = 6.0 (resonance not observed) and positive when
St, ---4-7 (resonance observed).
Figure l l ( a, b) shows sequences of "snapshot s", or i nst ant aneous pl ot s, highlighting
t he position of a large-scale vort ex st ruct ure as it passes across t he t op of t he
downst r eam baffle, for Sta = 4-7 and 6-0. It is useful in i nt erpret i ng t hese figures to
not e t hat for a vort ex travelling left to right, positive i nst ant aneous acoustic power P is
gener at ed if t here is a component of t he local acoustic particle velocity u upwards and
negative P if t here is a component of u downwards. For each Strouhal number , t he first
ACOUSTIC SOURCES IN A DUCT WITH BAFFLES 363
(c)
Fi gure 9. (continued)
"snapshot" is such that the acoustic particle velocity is zero and about to become
positive in the downstream direction. The following "snapshots" are separated in phase
angle by zc/2. The phase h (see equation (8)) of the acoustic cycle at which the vortex
arrives at the top of the second baffle has been stressed in the preceding description as
being an iml6ortant indicator of the sign and amount of acoustic power that is expected
to be generated by a vortex in its passage past the downstream baffle. A study of the
complete selection of "snapshots" of the flow, a sample of which are shown in Figure
l l (a, b), indicates that the phase offset h is approximately Jr for Sta =4. 7 and
approximately 0 for St a = 6.0. Figure 11 (a, b) also indicates the flow velocity and local
acoustic particle velocity at the vortex cloud centroid for the cases when the acoustic
particle velocity is nonzero.
The predicted history of acoustic power generation by a large-scale vortex structure
shed from the upstream baffle and passing the downstream set of baffles is shown in
~J
o
0
I I I I
A
i \
I I I I
- O 1 2 3 4 ~5
Ti me, x (acoustic cycles)
Fi gure 10. Predi ct ed i nst ant aneous acoust i c power P (arbitrary uni t s) gener at ed by t he ent i re flow at two
different acoustic St rouhal number s correspondi ng to a maxi mum and a mi ni mum of t he gener at ed acoustic
energy per cycle: - - , St a = 4-7 ( maxi mum) ; . . . . St, = 6-0 ( mi ni mum) .
(a)
Fi gure l l . Predi ct ed i nst ant aneous flow st ruct ures ( super i mposed f r om t he same phase of t he acoust i c
cycle over five consecut i ve acoust i c cycles) at quar t er acoust i c cycle i nt erval s for t he acoust i c St rouhal
number s (a) St,, = 4.7 and (b) St a = 6-0; ~, direction and relative magni t ude of vort ex cent roi d velocity;
(b)
---~, direction and relative magnitude of the acoustic particle velocity at the vortex centroid (scaled up by a
factor of 5). The vortex cloud under consideration has been highlighted by plotting a shaded circle at the
position of each of its constituent elemental vortices.
366 K . H O U R I G A N ET AL.
! 1
\ I
\ t
\ t
V
I
Fi gure 12. Pr edi ct ed i ns t ant aneous powe r out put ge ne r a t e d by a si ngl e l ar ge- scal e vor t ex passi ng t he
downs t r e a m baffle f or t wo acoust i c St r ouhal numbe r s : - - , St~ = 4-7; , Sta = 6.0. Th e di s t ance
f r om t he ups t r e a m baffle has be e n nor mal i zed wi t h r e s pe c t t o t he di s t ance b e t we e n t he baffl es.
Fi gur e 12 f or t he above t wo acoust i c St r ouhal number s of 4. 7 and 6.0, cor r es pondi ng t o
a local maxi mum and a l ocal mi ni mum, r espect i vel y, of t he acoust i c ener gy per cycl e
gener at ed by t he flow. The r es onant s ound field at t he assumed l evel det er mi nes t he
phase of t he acoust i c cycl e dur i ng whi ch a new vor t ex grows in t he s hear l ayers shed
f r om t he upst r eam set of baffles. Thi s phase is i ndependent of t he acoust i c St r ouhal
number ; as a resul t of t he angl e bet ween t he vor t ex vel oci t y and t he l ocal acoust i c
part i cl e vel oci t y, t he r egi on near t he ups t r eam baffle is a small acoust i c sink. Whe t he r
net posi t i ve ener gy is gener at ed over a l i fet i me of a vor t ex depends on t he phase of t he
acoust i c cycl e when t he vor t ex arri ves at t he downs t r eam baffle. For an acoust i c
St r ouhal numbe r of 4.7, a l arge-scal e vor t ex cont r i but es a net posi t i ve acoust i c ener gy
as it passes t he downs t r eam set of baffles ( r esonance obser ved) ; f or an acoust i c
St r ouhal numbe r of 6. 0, a net negat i ve acoust i c ener gy cont r i but i on resul t s whi ch
cor r es ponds t o no r esonance bei ng obser ved.
6. DI SCUSSI ON
The resul t s of Nomot o & Cul i ck (1982) and t he exper i ment s descr i bed in t hi s pa pe r
r ei nf or ce t he not i on t hat t he exi st ence of acoust i c r es onance depends on t he phase of
t he acoust i c cycl e at whi ch vort i ces, shed f r om t he upst r eam baffle each sound cycl e,
arri ve at t he downs t r eam baffle. In bot h sets of exper i ment s, no r egul ar vor t ex
sheddi ng was det ect ed when t he acoust i c r es onance was not exci t ed. Thi s is consi st ent
wi t h t he i dea t hat a phas e- dependent acoust i c sour ce can exist at a l ocat i on
downst r eam of t he vor t ex sheddi ng r egi on, as pr opos ed by St okes & Wel sh (1986) and
St oneman et al. (1988).
The resul t s of t he single vor t ex si mul at i on descr i bed her e (Fi gure 7) show how such a
phas e- dependent sour ce can exist. The phase of t he acoust i c cycl e at whi ch a vor t ex
arri ves at t he downs t r eam baffle is det er mi ned by t he phase offset h [see equat i on (8)].
That is, t he val ue of t he i nst ant aneous acoust i c power P gener at ed by a vor t ex is ver y
dependent on t he di r ect i on and ampl i t ude of t he acoust i c par t i cl e vel oci t i es when t he
vor t ex passes a gi ven poi nt . The net acoust i c ener gy gener at ed by t he vor t ex passi ng
t he baffle is equal t o t he i nt egral of P ove r t i me. For t he cases shown in Fi gur e 7(b) and
(d), wher e h = ~r/2 and h = 3Jr~2, r espect i vel y, t he vor t ex is passi ng over t he baffle
when t he local acoust i c par t i cl e vel oci t y is at a maxi mum and t he i nt egr al is zer o due t o
ant i symmet r y. That is, f or t he cases wher e t he acoust i c part i cl e vel oci t y is at a
maxi mum when t he vor t ex r eaches t he downs t r eam baffle, t he vor t ex gener at es no net
ACOUS T I C S OURCES I N A DUCT WI TH BAF F LES
367
acoustic energy over its time history. The vortex generates as much energy while
approaching the baffle as it absorbs while leaving the baffle. For h = Jr (Figure 7c)
maximum acoustic energy per cycle is generated. In this case, the vortex passes over
the baffle when the acoustic particle velocity is zero everywhere and is about to change
direction to upstream and oppose the mean flow. This value of the phase offset allows
sin [see equation (10)] to remain negative on both sides of the downstream baffle as
the vortex cuts across the acoustic particle velocity field lines leading to positive
contributions to the acoustic energy while both approaching and leaving the baffle. A
maximum net acoustic energy output results. For h = 0, the reverse is true. Here sin
remains positive during the traversal and energy absorption takes place both before
and after crossing the baffle.
In the single vortex model, the angle between the velocity direction of the vortex and
that of the local acoustic particle velocity are significant only in the region of the baffle.
That is, the vortex, which follows a horizontal path, cuts across the acoustic
"streamlines" near the baffle (shown in Figure 4(b)) to produce acoustic energy,
according to equations (9) and (10). This model explains the basic concepts behind the
acoustic energy generation process well but, importantly, it cannot predict the variation
of the phase offset h as a function of Strouhal number. The vortex cloud model enables
this to be done.
The results of the vortex cloud model support the concept of an acoustic source
which is phase-dependent. The shedding rate of the vortex clouds from the upstream
baffle is predicted to be locked to the sound frequency in the presence of the strong
sound field; the phase of the acoustic cycle at which these clouds begin to grow is
observed and predicted to be the same for different acoustic Strouhal numbers.
However, the phase of the acoustic cycle at which a cloud arrives at the downstream
baffle is strongly dependent on the acoustic Strouhal number (and therefore flow
velocity). Comparing the observed flows in Figure 6 with the predicted flows in Figure
9 shows that the vortex cloud model predicts satisfactorily the occurrence of different
numbers of vortices between the baffles at the appropriate locations for each of the
acoustic Strouhal numbers at which a peak sound pressure level is observed.
Furthermore, the acoustic Strouhal numbers (4.7 and 7.7) at which predicted peaks in
the acoustic energy per cycle is generated by the flow (Figure 8) correspond very
closely to those at which peaks in the sound pressure level are measured (Figure 5).
For Sta = 4-7, Figure l l (a) shows that the vortex cloud centre near the downstream
baffle moves from upstream of the baffle to downstream of the baffle as the acoustic
particle velocity changes from positive to negative (positive refers to the downstream
direction). Interpreted in terms of the single vortex model, the phase offset in this case
is h = ~r, corresponding to the case where maximum net acoustic energy is generated.
Experimentally, a peak in the sound pressure level was observed at this acoustic
Strouhal number (Figure 5).
A similar phenomenon occurs for the resonance occurring at an acoustic Strouhal
number of 7.7 at the correspondingly lower flow velocity. In this case, the vortex cloud
model also predicts that the vortex cloud passes over the downstream baffle
approximately when the acoustic particle velocity is changing sign from positive to
negative. This supports the predictions of the single vortex model, and confirms the
predicted phase relationship between the acoustic field and the arrival of the vortex
approaching the downstream baffle necessary to maximize the production of net
acoustic energy and therefore sound pressure level.
The instantaneous acoustic power generated for Sta = 4-7 is found to have two
positive peaks each sound cycle (Figure 12), as predicted also by the single vortex
368 K. HOURI GAN ET AL.
model in Fi gur e 7(c). Thi s resul t s in a net posi t i ve acoust i c ener gy bei ng gener at ed per
acoust i c cycle at this acoust i c St r ouhal number , as seen in Fi gur e 8. Howe ve r , in
cont r ast wi t h t he single vor t ex model resul t s, t he peaks ar e not equal in magni t ude.
Mor e det ai l ed analysis and vi ewi ng of vi deot ape r ecor di ngs of t he dynami c si mul at i on
r eveal t hat , as a resul t of compl ex i nt er act i on bet ween t he vor t ex cl oud and t he
downst r eam baffle and loss of vort i ci t y i nt o t he cavi t y, t he net ci r cul at i on of t he vor t ex
cl oud is decr eased as it passes t he downs t r eam baffle. Cons equent l y, t he second pe a k is
not as subst ant i al as t he first. Nonet hel ess, al t hough slightly modi f i ed in det ai l , t he
single vor t ex model pr edi ct i on of posi t i ve cont r i but i ons f r om bot h sides of t he baffle in
t he r esonant si t uat i on is bor ne out by t he mor e sophi st i cat ed model .
For Sta = 6. 0, t he vor t ex cl oud arri ves at t he downst r eam baffle hal f a sound cycl e
l at er [Fi gure l l ( b) ] t han f or t he case wher e Sta = 4.7, resul t i ng in a net negat i ve
gener at i on of acoust i c ener gy (Fi gure 12). The acoust i c r es onance woul d not be
sust ai ned under t hese ci rcumst ances, whi ch is consi st ent wi t h t he obs er vat i on of no
r ecor ded r esonance at this acoust i c St r ouhal number . Not i ce t hat f or this St r ouhal
number (6-0) t he second negat i ve peak in Fi gur e 12 ( af t er t he vor t ex cl oud has passed
t he second baffle) is al most nonexi st ent , whi ch di ffers f r om t he resul t of t he single poi nt
vor t ex model (Fi gure 7( a) ) . The di f f er ence seems t o be due in par t t o t he a mount of
ci rcul at i on r et ai ned by t he vor t ex as it passes t he downs t r eam baffle; t he ci rcul at i on is
unchanged in t he poi nt vor t ex model , wher eas it decr eases in t he vor t ex cl oud model .
Some of this ci rcul at i on decr ease resul t s f r om t he sound field pushi ng par t of t he vor t ex
cl oud i nt o t he cavi t y as it appr oaches t he downs t r eam baffle. Cons equent l y, much less
of t he t ot al ci rcul at i on passes over t he baffle at this St r ouhal numbe r r el at i ve t o t hat f or
t he r esonant St r ouhal numbe r l eadi ng t o less absor pt i on of acoust i c ener gy t han woul d
ot her wi se occur. Howe ve r , t he net sour ce of acoust i c ener gy is still negat i ve i ndi cat i ng
t hat t he hypot het i cal l oud acoust i c r es onance assumed in t he model cannot be
sust ai ned at this acoust i c St r ouhal number . Thi s means t hat t he sound pr essur e l evel
woul d not bui l d up t o l oud r esonant l evel s in pr act i ce, as veri fi ed by t he exper i ment s.
The l i near stability t heor y of Fl andr o (1986) also i ndi cat es t hat cer t ai n r es onant modes
of a r ocket chamber will grow in ampl i t ude in onl y cer t ai n r anges of t he St r ouhal
number .
It is t her ef or e pr opos ed t hat t he i nfl uence of t he r esonant sound field in synchroni s-
ing t he vor t ex sheddi ng is a vi t al par t of t he over al l r es onance process. Thi s is
effect i vel y acoust i c f eedback. Nomot o & Cul i ck (1982) cast doubt on t he r ol e of
f eedback, saying t hat t he di st ance f r om sour ce t o f eedback site is t oo small f or
compressi bi l i t y t o be i mpor t ant . Thi s di st ance, however , is not t he one t hat mat t er s.
The f eedback is via a r esonance, whi ch means t hat most of t he sound ener gy whi ch
i nfl uences vor t ex sheddi ng at any one t i me was gener at ed in pr evi ous cycles, and has
been r ef l ect ed, usual l y many t i mes, f r om t he duct t er mi nat i ons.
Fi nal l y, it is not ed t hat t he pr esent i nvest i gat i on reveal s an acoust i c p h e n o me n o n
whi ch is f undament al l y similar t o t he case of flow past a squar e l eadi ng edge pl at e
( St okes & Wel sh 1986). In each case, t he vor t ex sheddi ng poi nt ( upst r eam baffle or
squar e l eadi ng edge of pl at e) is r e mot e f r om a downs t r eam acoust i c sour ce r egi on
( downst r eam baffle or trailing edge of pl at e) . The acoust i c r esonance can be sust ai ned
onl y when t he vort i ces, shed at t he acoust i c f r equency via f eedback of t he r es onant
acoust i c mode, arri ve at t he downs t r eam sour ce r egi on at a par t i cul ar phase of t he
acoust i c cycle. The phase of arri val of a vor t ex at t he downs t r eam sour ce r egi on is a
f unct i on of t he flow vel oci t y; t he par t i cul ar phase r equi r ed f or acoust i c ener gy t o be
gener at ed can occur f or a numbe r of di f f er ent flow vel oci t i es. The r e f or e , acoust i c
r esonance is obser ved over di scret e vel oci t y r anges in bot h cases even t hough t he pl at e
ACOUSTIC SOURCES IN A DUCT WITH BAFFLES 369
and baffle geomet r i es ar e di f f er ent and t he pr es ent r e s ona nt acoust i c mo d e is
pri nci pal l y l ongi t udi nal r a t he r t han t r ansver se.
7. CONCL US I ONS
An essent i al l y l ongi t udi nal r es onant acoust i c mo d e can be exci t ed in a duct by fl ow
over t wo set s of baffles. Pe a k s ound pr es s ur e l evel s of t hi s mo d e ar e obs e r ve d t o occur
when l arge-scal e vor t i ces, f or me d in t he shear l ayer s s epar at i ng f r om t he ups t r e a m set
of baffles, a ppr oa c h t he downs t r e a m set of baffl es at a par t i cul ar phas e of t he i nduced
resonant acoust i c cycl e.
I ncor por at i on of Ho we ' s (1975) t he or y of r es onant s ound i nt o t wo di f f er ent fl ow
model s enabl es t he de t e r mi na t i on of t he sound sour ces in t he fl ow and an expl anat i on
of t he occur r ence of p e a k s ound pr essur e l evel s at par t i cul ar fl ow vel oci t i es.
The f e e dba c k of t he s ound field, vi a a vel oci t y pe r t ur ba t i on, on t he s epar at i ng shear
l ayer f r om t he ups t r e a m baffl e synchr oni zes t he vor t ex s heddi ng t o t he s ound, r esul t i ng
in vort i ces bei ng f or me d at t he s ame i nst ant of t he acoust i c cycl e i ndependent l y of t he
acoustic St r ouhal numbe r . The gener at i on of acoust i c e ne r gy neces s ar y t o sust ai n an
acoustic r es onance has be e n s hown t o de pe nd on t he phas e of t he acoust i c cycl e at
which a vor t ex passes t he downs t r e a m baffle. Thi s phas e is a f unct i on of t he t i me t aken
for a vor t ex t o r each t he downs t r e a m baffle, whi ch in t ur n de pe nds on t he acoust i c
St rouhal number . Thi s expl ai ns why acoust i c r es onance can onl y be sust ai ned f or
part i cul ar acoust i c St r ouhal numbe r s .
ACKNOWL E DGE ME NT
The aut hor s wi sh t o t ha nk Mr N. B. Ha mi l t on f or phot ogr a phi ng t he fl ows descr i bed in
t he paper .
REFERENCES
ARCHIBALD, F. S. 1975 Self-excitation of an acoustic resonance by vortex shedding. Journal of
Sound and Vibration 38, 81-103.
BROWN, R. S., DUNLAP, R. , YOUNG, S. W. & WAUGH, B. G. 1981 Vort ex shedding as a source
of acoustic energy in segmented solid rockets, fournal of Spacecraft and Rockets 18,
312-319.
BLEVINS, R. D. 1985 The effect of sound on vortex shedding from cylinders. Journal of Fluid
Mechanics 61, 217-237.
CHUHG, T. J. & SOHN, J. L. 1986 Interactions of coupled acoustic and vortical instability. AI AA
Journal 24, 1582-1596.
CUMPSTY, N. A. & WHITEHEAD, D. S. 1971 The excitation of acoustic resonances by vortex
shedding. Journal of Sound and Vibration 18, 353-369.
FLANDRO, G. A. 1986 Vortex driving mechanism in oscillatory flows. Journal of Propulsion and
Power 2, 206-214.
HARRIS, R. E. , WEAVER, D. S. & DOKAINISH, M. A. 1988 Unstable shear layer oscillations past
a cavity in air and water flows. In Proceedings of 1988 International Symposium on
Flow-Induced Vibration and Noise; Vol. 6: Acoustic Phenomena and Interaction in Shear
Flows over Compliant and Vibrating Surfaces (eds W. L. Keith, E. M. Ur am & A. J.
Kalinowski), pp. 13-24. New York: ASME.
HEGDE, U. C. & STRAHLE, W. C. 1985 Sound generation by turbulence in simulated rocket
mot or cavities. AIAA Journal 23, 71-77.
HowE, M. S. 1975. Contributions to the theory of aerodynamic sound, with application to excess
jet noise and the theory of the flute. Journal of Fluid Mechanics 71, 625-673.
HowE, M. S. 1980 The dissipation of sound at an edge. Journal of Sound and Vibration 70,
407-411.
370 K. HOURIGAN ET AL.
JOHNSON, C. O. tg~LLOERHRKE~ R. I. 1984 An experimental investigation of wake edge tones.
AI AA Journal 22, 1249-1253.
LEwis, R. I. 1981 Surface vorficity modelling of separated flows from two-dimensional bluff
bodies of arbitrary shape. Journal of Mechanical Engineering Science 23, 1-23.
NOMOTO, H. & CULICK, E. C. 1982 An experimental investigation of pure tone generation by
vortex shedding in a duct. Journal of Sound and Vibration 84, 247-252.
PARKEa, R. 1966 Resonant effects in wake shedding from parallel plates: some experimental
observations. Journal of Sound and Vibration 4, 62-72.
PARKER, R. 1967 Resonant effects in wake shedding from parallel plates: calculation of resonant
frequencies. Journal of Sound and Vibration 5, 330-343.
PARKER, R. 1969 Discrete frequency noise generation due to flow over blades, supporting
spokes and similar bodies. ASME, Paper No. 69-WA/GF13.
PARKER, R. GRIFFITHS, W. M. 1968 LOW frequency resonant effects in wake shedding from
parallel plates. Journal of Sound and Vibration 7, 371-379.
SMITH, P. A. & STANSBY, P. K. 1988 Impulsively started flow around a circular cylinder by the
vortex method. Journal of Fluid Mechanics 194, 45-77.
STorms, A. N. & WELSH, M. C. 1986 Flow-resonant sound interaction in a duct containing a
plate, II: square leading edge. Journal of Sound and Vibration 104, 55-73.
STONEMAN, S. A. T., HOURI(3AN, K., STOKES, A. N. & WELSH, M. C. 1988 Resonant sound
caused by flow past two plates in tandem in a duct. Journal of Fluid Mechanics 192,
455-484.
WELSH, M. C. & GmsoN, D. C. 1979 Interaction of induced sound with flow past a square
leading edged plate in a duct. Journal of Sound and Vibration 67, 501-511.
WELSH, M. C., STOKES, A. N. t~ PARKER, R. 1984 Flow-resonant sound interaction in a duct
containing a plate, Part I: semi-circular leading edge. Journal of Sound and Vibration 95,
305-323.
WELSH, M. C., HOURIGAN, K., WELCH, L. W., DOWNIE, R. J., TrIOMeSON, M. C. & STOKES,
A. N. 1990. Acoustic and experimental methods: the influence of sound on flow and heat
transfer. Experimental Thermal and Fluid Science 3, 138-152.
WARSL Z. U. A. 1982 Basic differential models for coordinate generation. In Numerical Grid
Generation, (ed. J. F. Thompson), pp. 41-77. Amsterdam: North-Holland.
ZIENKIEWICZ, O. C. 1978 The Finite Element Method. New York: McGraw-Hill.